Add modules= config option for ResourcesBot. (#127)

Allow default post-processing modules to be set in config.ini while
keeping -m as a command-line override.

Co-authored-by: Cursor <cursoragent@cursor.com>

Author: holybiber

config.example.ini

@@ -56,6 +56,10 @@ username = ResourcesBot
 password = MySecretPassword
 
 # resourcesbot writes language reports to [Paths:languagereports]
+# Post-processing modules to run (abbreviations, space-separated).
+# Use "all" to run every module (default). Overridden by -m on the command line.
+# Run `python3 resourcesbot.py -h` to see available module abbreviations.
+modules = all
 # Optionally: log to files (will be relative to path defined in [Paths:logs] )
 # Three different verbosity levels (warning, info, debug)
 logfile = resourcesbot.log

pywikitools/resourcesbot/bot.py

@@ -57,6 +57,43 @@ def load_module(module_name: str) -> Callable:
     raise RuntimeError(f"Couldn't load module {module_name}. Giving up")
 
 
+def build_module_choices() -> Dict[str, str]:
+    """Return module abbreviations mapped to full module names."""
+    modules: Dict[str, str] = {}
+    for selected_module in AVAILABLE_MODULES:
+        module = load_module(selected_module)
+        modules[module.abbreviation()] = selected_module
+    return modules
+
+
+def resolve_run_modules(
+    modules: Dict[str, str],
+    config: ConfigParser,
+    cli_modules: Optional[List[str]] = None,
+) -> list[str]:
+    """
+    Resolve which post-processing modules to run.
+
+    Command-line ``-m`` overrides ``[resourcesbot] modules`` in config.ini.
+    """
+    if cli_modules is not None:
+        return [modules[abbr] for abbr in cli_modules]
+
+    config_value = config.get("resourcesbot", "modules", fallback="all").strip()
+    if config_value.lower() == "all":
+        return AVAILABLE_MODULES
+
+    selected = config_value.split()
+    unknown = [abbr for abbr in selected if abbr not in modules]
+    if unknown:
+        available = ", ".join(sorted(modules.keys()))
+        raise ValueError(
+            "Unknown module(s) in config.ini [resourcesbot] modules: "
+            f"{', '.join(unknown)}. Available: {available}"
+        )
+    return [modules[abbr] for abbr in selected]
+
+
 class ResourcesBot:
     """Contains all the logic of our bot"""
 

pywikitools/test/test_resourcesbot.py

@@ -14,7 +14,13 @@
 
 import pywikibot
 
-from pywikitools.resourcesbot.bot import ResourcesBot, load_module
+from pywikitools.resourcesbot.bot import (
+    AVAILABLE_MODULES,
+    ResourcesBot,
+    build_module_choices,
+    load_module,
+    resolve_run_modules,
+)
 from pywikitools.resourcesbot.data_structures import TranslationProgress, WorksheetInfo
 from pywikitools.test.test_data_structures import TEST_PROGRESS, TEST_TIME, TEST_URL
 
@@ -341,5 +347,53 @@ def factory(*args, **kwargs):
     # TODO: test_run_with_limit_lang
 
 
+class TestResolveRunModules(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        cls.modules = build_module_choices()
+
+    def _config_with_modules(self, modules_value: str) -> ConfigParser:
+        config = ConfigParser()
+        config.read_dict(
+            {
+                "resourcesbot": {
+                    "site": "test",
+                    "username": "TestBotName",
+                    "modules": modules_value,
+                }
+            }
+        )
+        return config
+
+    def test_default_is_all_modules(self):
+        config = ConfigParser()
+        config.read_dict({"resourcesbot": {"site": "test", "username": "TestBotName"}})
+        self.assertEqual(resolve_run_modules(self.modules, config), AVAILABLE_MODULES)
+
+    def test_config_all(self):
+        config = self._config_with_modules("all")
+        self.assertEqual(resolve_run_modules(self.modules, config), AVAILABLE_MODULES)
+
+    def test_config_specific_modules(self):
+        config = self._config_with_modules("list report")
+        self.assertEqual(
+            resolve_run_modules(self.modules, config),
+            ["write_lists", "write_report"],
+        )
+
+    def test_cli_overrides_config(self):
+        config = self._config_with_modules("list")
+        self.assertEqual(
+            resolve_run_modules(self.modules, config, cli_modules=["report"]),
+            ["write_report"],
+        )
+
+    def test_config_unknown_module_raises(self):
+        config = self._config_with_modules("list unknown")
+        with self.assertRaises(ValueError) as context:
+            resolve_run_modules(self.modules, config)
+        self.assertIn("unknown", str(context.exception))
+
+
 if __name__ == "__main__":
     unittest.main()

resourcesbot.py

@@ -65,9 +65,14 @@
 import sys
 import traceback
 from configparser import ConfigParser
-from typing import Dict, List
+from typing import List
 
-from pywikitools.resourcesbot.bot import AVAILABLE_MODULES, ResourcesBot, load_module
+from pywikitools.resourcesbot.bot import (
+    ResourcesBot,
+    build_module_choices,
+    load_module,
+    resolve_run_modules,
+)
 
 
 def parse_arguments() -> ResourcesBot:
@@ -85,16 +90,17 @@ def parse_arguments() -> ResourcesBot:
 
     log_levels: List[str] = ["debug", "info", "warning", "error"]
     rewrite_options: List[str] = ["all", "json", "summary"]
-    modules: Dict[str, str] = {}  # abbreviation -> full name
+    modules = build_module_choices()
     modules_help = "Select the modules to be run. Available options are:\n"
-    # Read module information from the module classes
-    for selected_module in AVAILABLE_MODULES:
+    for abbr, selected_module in modules.items():
         module = load_module(selected_module)
-        modules_help += f" - {module.abbreviation()}: {module.help_summary()}\n"
-        modules[module.abbreviation()] = selected_module
+        modules_help += f" - {abbr}: {module.help_summary()}\n"
         if module.can_be_rewritten():
-            rewrite_options.append(module.abbreviation())
-    modules_help += "Default: run all modules"
+            rewrite_options.append(abbr)
+    modules_help += (
+        "Default: all modules, or set [resourcesbot] modules= in config.ini "
+        "(overridden by -m)"
+    )
 
     parser.add_argument(
         "--read-from-cache",
@@ -133,11 +139,7 @@ def parse_arguments() -> ResourcesBot:
     assert isinstance(numeric_level, int)
     set_loglevel(config, numeric_level)
 
-    # Map abbreviations to full module names
-    if args.m is None:
-        run_modules = AVAILABLE_MODULES
-    else:
-        run_modules = [modules[abbr] for abbr in args.m]
+    run_modules = resolve_run_modules(modules, config, args.m)
 
     return ResourcesBot(
         config=config,